Universal Agent Skills
Back to Collection
Skill/Orchestration/Methodology Inspired

LLM Council

Five advisor debate protocol peer reviewing proposals anonymously.

What It Does

Runs complex decisions through five independent agent perspectives, executes anonymous peer review rounds, and synthesizes a balanced verdict accounting for disagreements.

Origin and Adaptation Notes

Individual model answers frequently convey unearned confidence. Anonymous review rounds surface vulnerabilities and trade offs.

Attribution: Protocol implementation inspired by Andrej Karpathy's LLM Council methodology.

When to Activate

  • •Evaluating critical architectural decisions and tech stack tradeoffs
  • •Stress testing project proposals before committing engineering resources
  • •Surfacing subtle failure modes across complex decisions

When to Avoid

  • •Simple factual lookups or basic binary questions

Skill Specification and Runtime Setup

Full Instruction Specification
# LLM Council Specification

## Overview
Runs complex decisions through five independent agent perspectives, executes anonymous peer review rounds, and synthesizes a balanced verdict accounting for disagreements.

## Rationale
Individual model answers frequently convey unearned confidence. Anonymous review rounds surface vulnerabilities and trade offs.

## Core Principles
* Always verify environment constraints and dependencies before executing.
* Enforce negative constraints and reject speculative abstractions.
* Validate output against stated acceptance rubrics.

## Execution Workflow
1. Discovery: Map requirements and inspect active context.
2. Decomposition: Break down the task into discrete, verifiable phases.
3. Execution: Implement surgical changes preserving existing code architecture.
4. Verification: Test all criteria before declaring completion.

Related Skills in Orchestration